To ensure that the most important types of HNSCC tumors are adequately represented, the laryngeal tumor-based cell line HLaC78 and the tongue tumor-based cell line Cal27 were chosen. The HNSCC cell line HLaC78 was isolated from a larynx carcinoma of a male patient by Professor Hans-Peter Zenner in the Department of Oto-Rhino-Laryngology, Plastic, Aesthetic and Reconstructive Head and Neck Surgery of the University Hospital (Würzburg, Germany) (RRID: CVCL_6647) (16). The Cal27 cell line was first isolated from the tongue tumor of a 56-year-old male patient (17), which was purchased at American Type Culture Collection. The cells were cultured in RPMI-1640 (Biochrom, Ltd.) containing 10% fetal calf serum (FCS; Linaris Biologische Produkte GmbH), 1% penicillin and streptomycin (Sigma-Aldrich, Merck KGaA) at 37°C with 5% CO₂. The medium was changed every 2 days. After reaching 70–80% confluence, cells were detached by trypsinization with 0.25% trypsin (Gibco; Thermo Fisher Scientific, Inc.), washed with PBS, counted before 1×10⁶ cells were seeded into new 250-ml culture flasks. Cells in the exponential growth phase were used for subsequent experiments.

Bone marrow was donated by 10 voluntary patients (5 male and 5 female; mean age 63.2 years), who had undergone surgery in the Department of Orthopedics, Koenig-Ludwig-Haus (University Hospital Würzburg, Germany). All patients agreed by providing written informed consent. The present study was approved by the Ethics Committee of the Medical Faculty of the University of Würzburg (approval no. 91/19). Bone marrow was harvested from acetabular reaming material as waste material from patients undergoing total hip arthroplasty surgery at the Department of Orthopedic Surgery, under aseptic conditions, and patients with clinical signs of osteoporosis, cancer or infectious disease were excluded. hMSCs were isolated in accordance with the protocol of Lee et al (18), which was also described in detail previously (19). Briefly, hMSCs were isolated by Ficoll (density=1.077 g/ml; Biochrom, Ltd) density gradient centrifugation (30 min; at room temperature; 800 × g; brake and acceleration levels set to the lowest value). After centrifugation, a clear phase separation was observed with a clearly visible optical dense interphase containing mononuclear cells. Cells from this interphase were pipetted in a new 50-ml reaction tube and subsequently washed with PBS twice. Cell culture was performed in the expansion medium (DMEM-EM), which consisted of DMEM (Gibco; Thermo Fisher Scientific, Inc.) containing 4.5 g/l D-Glucose, 10% FCS (Linaris Biologische Produkte GmbH), 1% penicillin and streptomycin (Sigma-Aldrich; Merck KGaA), whereas incubation was at 37°C and 5% CO₂. hMSC morphology was evaluated by capturing phase contrast images using an inverted light microscope at 100× magnification (DMI 4000b Inverted Microscope, Leica Microsystems GmbH).

According to the guidelines provided by the International Society of Cellular Therapy (ISCT), hMSC should be adherent to plastic surfaces and positive for the expression of surface markers CD105, CD90 and CD73 but negative for the expression of hematopoietic surface markers, including CD45 or CD34 (20,21). Furthermore, hMSC should demonstrate multipotency in vitro (20,21). Plastic adherence was assessed using inverted light microscopy at ×10-40 magnification (Leica DMI 4000b Inverted Microscope; Leica Microsystems GmbH). hMSC surface marker expression was evaluated by flow cytometry. After detachment, cells were washed with PBS and cultured with 5% FCS on ice for 1 h. Afterwards, hMSCs (1×10⁶) were incubated with anti-CD90 (dilution 1:500; conjugate APC; cat. no. 559869; BD Biosciences), anti-CD73 (dilution 1:50; conjugate PE; cat. no. 550257; BD Biosciences), anti-CD45 (dilution 1:50; conjugate FITC; cat. no. 555482; BD Biosciences), anti-CD44 (dilution 1:50; conjugate FITC; cat. no. 555478; BD Biosciences) and anti-CD34 (dilution 1:50; conjugate PE; cat. no. 550761; BD Biosciences) antibodies for 1 h at 4°C and flow cytometric analysis was performed (BD FACSCanto™; BD Biosciences) and further analyzed by FACS Diva Software v5.0.3 (BD Biosciences).

The pluripotency of hMSCs was evaluated by staining. First, hMSC were cultured in osteogenic and adipogenic media. hMSC control was cultured in DMEM-EM medium. The osteogenic differentiation medium was comprised of DMEM-EM, supplemented with 10⁻⁷ M dexamethasone, 10⁻³ M β-glycerophosphate and 10⁻⁴ M ascorbate-2-phosphate (all Sigma-Aldrich; Merck KGaA). The adipogenic differentiation medium was comprised of DMEM-EM, combined with 10⁻⁷ M dexamethasone and 10⁻⁹ g/ml recombinant human insulin (Sigma-Aldrich; Merck KGaA). hMSCs incubated with the osteogenic medium were termed osteo-hMSCs whereas hMSCs incubated with the adipogenic medium were termed adipo-hMSC at 37°C with 5% CO₂ for 3 weeks.

For the evaluation of the osteogenic differentiation, von Kossa and Alizarin-Red staining were performed to detect calcium mineral components. For von Kossa staining the cells were first washed with distilled water, incubated with 1% silver nitrate solution at room temperature (diluted in distilled water; cat. no. #7908.1; Carl Roth) under UV-light for 20 min, washed three times with distilled water, incubated with 5% sodium thiosulfate pentahydrate (diluted in distilled water; cat. no. #6516.0500; Merck KGaA), washed three times with distilled water, incubated with Nuclear Fast Red solution for 5 min at room temperature [5 g Aluminum sulfate hydrate (cat. no. #227617; Sigma-Aldrich; Merck KGaA) in 100 ml distilled water, 0.1 g Nuclear Fast Red (cat. no. #5188; Sigma-Aldrich; Merck KGaA)], washed three times with distilled water, incubated with ascending alcohol series and dried for microscopy. The Alizarin-Red stock solution was prepared by dissolving 2 g of Alizarin-Red S (cat. no. #K00332679; Merck KGaA) in 100 ml distilled water. The pH-value was adjusted at 4.1-4.3 by adding of glacial acetic acid (cat. no. #1000661000; Merck KGaA). For staining the cells were incubated with the stock solution for 5 min at room temperature. Before and after incubation the cells were washed with distilled water. Adipogenic differentiation was assessed using Oil Red O staining to reveal intracellular lipid droplets. For preparing of the Oil Red O staining stock solution 0.5 g Oil Red O (cat. no. #O0625; Sigma-Aldrich; Merck KGaA) was dissolved in 100 ml Propylenglycol (cat. no. #P4347; Sigma-Aldrich; Merck KGaA) at 60°C. For the staining procedure the cells was washed with distilled water, incubated with Propylenglycol for 5 min at room temperature, then with 60°C warm Oil Red O stock solution for 10 min, washed with Propylenglycol, washed three times with distilled water and stained with Mayers Hematoxylin-solution (cat. no. #1.09249; Merck KGaA) for 30 sec. Until microscopy the cells were covered with PBS. All images were acquired with a light microscope (DMI 4000b Inverted Microscope; Leica Microsystems GmbH).

RT-qPCR was used to verify hMSC differentiation. The following primers were chosen for osteogenic differentiation: i) Alkaline phosphatase (ALPL; cat. no. 4331182; assay ID, Hs01029144_m1); ii) osteocalcin (BGLP; cat. no. 4331182; assay ID, Hs01587814_g1); iii) collagen 1 (Col 1; cat. no. 4331182; assay ID, Hs0016004_m1); and iv) runt-related transcription factor 2 (RUNX-2; cat. no. 4331182; assay ID, Hs00231692_m1). The following primers were chosen for adipogenic differentiation: i) fatty acid binding protein 4 (FABP4; cat. no. 4331182; assay ID, Hs01086177_m1); ii) leptin (LEP; cat. no. 4331182; assay ID, Hs00174877_m1); and iii) lipoprotein lipase (LPL; cat. no. 4331182; assay ID, Hs00173425_m1). GAPDH (cat. no. 4331182; assay ID, Hs02758991_g1) was used as the housekeeping gene. All primers were purchased from Thermo Fisher Scientific, Inc., the primer sequences of which are not publicly available. RT-qPCR was performed as follows: For total RNA extraction from hMSCs an RNeasy Kit (Qiagen GmbH) was used according to the manufacturer's protocol. For reverse transcription, the isolated RNA was converted into cDNA using SuperScript™ VILO™ Master Mix (cat. no. #11755-500; Invitrogen; Thermo Fisher Scientific, Inc.). The following temperature protocol was used for reverse transcription: 25°C for 10 min; 42°C for 59 min; 85°C for 5 min; 4°C for 2 min. Subsequent qPCR was performed using SYBR Green Real-Time PCR Master Mix (Thermo Fisher Scientific, Inc.) in a StepOnePlus™ thermocycler system (Applied Biosystems; Thermo Fisher Scientific, Inc.). The first denaturation step was 10 min at 95°C. Afterwards the following thermocycling protocol was utilized for 40 cycles: 50°C for 2 min, 60°C for 1 min and 95°C for 15 sec. The 2^−ΔΔCq method was applied to quantify the relative gene expression levels (22). The gene expression values are then normalized to those on of the hMSC control.

The co-culture experiments were performed in Transwell systems with a polyester membrane and pore size of 0.4 µm (Costar^® Transwell^®; Corning, Inc.). After seeding 60,000 hMSCs into the lower chambers of 12-well plates in DMEM-EM medium and microscopic confirmation of adherence, 60,000 HLaC78 and 60,000 Cal27 cells were seeded onto the Transwell inserts in DMEM-EM medium. hMSC without co-cultivation served as the control. Cells were kept in this co-culture system for 3 weeks at 37°C and 5% CO₂. The DMEM-EM medium was changed every 2 days. After a period of 21 days, hMSC differentiation into osteogenic and adipogenic lineages were determined using staining and RT-qPCR. This experiment was repeated 10 times using hMSCs from all 10 different donors.

The viability of HLaC78 and Cal27 cells co-cultured with hMSCs, adipo-hMSCs or osteo-hMSCs was measured by counting the cells using an electronic cell counter (CASY Cell Counter; OMNI Life Science GmbH). Identical measurements were also performed after the cultivation of the two HNSCC tumor cell lines with the conditioned medium of hMSCs (hMSC-CM), adipo-hMSC (adipo-hMSC-CM) and osteo-hMSC (osteo-hMSC-CM) at 37°C for 3 days. Conditioned media were obtained after 3 days at 37°C of hMSC, adipo-hMSC and osteo-hMSC incubation with DMEM-EM. Before the conditioning process the differentiation media was removed.

The Human Cytokine Array C3 dot blot assay (cat. no. AAH-CYT-3-4; Raybiotech, Inc.) was used to measure hMSC, adipo-hMSC and osteo-hMSC cytokine secretion after incubation with their respective differentiation media for 3 weeks. After removing of the differentiation media hMSCs, adipo-hMSCs and osteo-hMSCs were first incubated with DMEM without any supplements. After a period of 48 h at 37°C, the supernatants of the hMSCs from the 10 patients were then collected and pooled. The cytokine profile was analyzed according to the manufacturer's protocol. The chemiluminescence was assessed using an X-ray film. Semi-quantitative detection of IL-6 concentration was performed by density measurements using the ImageJ software (version 1.52a; National Institutes of Health) in relation to the positive control dot density. According to the manufacturer's declarations, the signal of the positive control spots is associated with the amount of biotinylated antibody printed onto the array.

Supernatants of the Cal27 cell culture after co-culture with hMSCs or incubation with the hMSC-CM for 3 days at 37°C were collected and analyzed for IL-6 levels using the ELISA kit human IL-6 (cat. no. 950.030.096; Diaclone SAS). All experiments were repeated with hMSCs from seven donors. The plates were read out at 450 nm (Titertek Multiskan PLUS; Thermo Fisher Scientific, Inc.). The standard curve was created by recombinant IL-6.

Cal27 and HlaC78 cells were incubated with hMSC-CM with or without 5 µg/ml anti-IL6 (cat. No. MAB2061; R&D Systems, Inc.), adipo-hMSC-CM and osteo-hMSC-CM at 37°C for 2 days. After washing the Cal27 and HlaC78 cells with PBS, they were harvested using a cell scraper and dissolved in RIPA buffer (PBS containing 1% NP40, 0.5% sodium deoxycholate and 0.1% SDS) supplemented with 10 µg/ml phenylmethanesulfonyl fluoride. Protein determination was performed by bicinchoninic acid method (Pierce BCA Protein Assay Kit; cat. no. #23227; Thermo Fisher Scientific, Inc.) Equal amounts (20 µg) of the total protein lysates were separated in a 10% SDS-polyacrylamide gel, before they were transferred onto polyvinylidene difluoride membranes. The membranes were blocked for 1 h at room temperature with TBST (10 mM Tris, 150 mM NaCl and 0.05% Tween-20, pH 8.0) containing 5% non-fat dry milk. The membranes were next incubated with the primary antibodies against phosphorylated (p-) STAT3 (1:1,000; rabbit; cat. No. 9145; Cell Signaling Technology, Inc.), STAT3 (1:2,000; rabbit; cat. No. 12640; Cell Signaling Technology, Inc.) and β-actin (1:2,000; mouse; cat. No. MA5-15739; Thermo Fisher Scientific, Inc.) overnight at 4°C. The membranes were then washed with TBST (Tween 0.1%) and incubated with a species-specific HRP-conjugated IgG secondary antibody (1:10,000; cat. no. 7074; Cell Signaling Technology, Inc.) for 1 h at room temperature. The bands were visualized using a chemiluminescence system (iBright1500; Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol.

All data were transferred into standard spreadsheets. Differences between groups were examined for significance, with one-way ANOVA performed using GraphPad Prism 6.0 statistics software (GraphPad Software, Inc.). For post hoc testing Dunnett´s multiple comparisons test was used (Fig. 3), for multiple comparisons Tukey's test was used (Fig. 5, Fig. 6, Fig. 7). All results were presented as mean ± SD. P<0.05 was considered to indicate a statistically significant difference and marked with an asterisk.

The hMSCs exhibited a fibroblast-shaped morphology when observed using microscopy (Fig. 1A). The Oil Red O, von Kossa and Alizarin Red staining revealed characteristics of osteogenic and adipogenic differentiation in osteo-hMSCs and adipo-hMSCs, respectively (Fig. 1). The successful osteogenic and adipogenic differentiation of hMSC was verified by qPCR.

To evaluate the extent of hMSC differentiation towards the osteogenic and adipogenic lineages, RT-qPCR was performed. hMSCs incubated in DMEM-EM without differentiation medium served as the control. After 1 week of incubation with either osteogenic or adipogenic media, the expression of adipogenic cell markers FABP4, LEP and LPL and osteogenic cell markers ALPL, BGLP, Col 1 and RUNX-2 were measured. Compared with that in the control group, adipogenic differentiation medium induced a 229.4-fold increase in FABP4 expression, a 275.4-fold increase of LPL expression and a 206.6-fold increase of LEP expression expression. In terms of osteogenic differentiation, there was a 4.2-fold increase in ALPL expression, a 3.7-fold increase in BGLP expression, a 1.5-fold increase in Col 1 expression and a 1.4-fold increase in RUNX2 expression (Table I).

According to flow cytometry analysis, the hMSCs were found to express surface markers CD90, CD73 and CD44 (Fig. 2). By contrast, hematopoietic markers CD45 and CD34 could not be detected (Fig. 2).

Co-culturing hMSCs with Cal27 or HLaC78 increased the expression of osteogenic and adipogenic lineages markers measured by RT-qPCR, compared with that in control cells (Fig. 3). However there was only a slightly increase of adipogenic markers after co-culturing hMSCs with Cal27. Furthermore, compared with that in the control group, Oil Red O staining of hMSCs co-cultured with Cal27 and HLaC78 cells showed markedly higher lipid droplet production (Fig. 4A-C). According to von Kossa staining, the quantity of calcium deposits was only increased slightly after hMSC co-cultivation with Cal27, but was more notably increased after co-cultivation with HLaC78 cells (Fig. 4D-F).

After co-cultivation of hMSC, adipo-hMSC and osteo-hMSC with Cal27 and HLaC78 cells, the number of HNSCC cells were counted. The number of Cal27 cells was increased significantly after co-cultivation with adipo-hMSC and osteo-hMSC compared with that in the groups of Cal27 cells that were not co-cultured (Fig. 5). Furthermore, there was an increased Cal27 cell count after co-cultivation with adipo-hMSC compared with hMSC (Fig. 5). In addition, the count of viable HLaC78 cells was significantly higher after co-cultivation with adipo-hMSC (Fig. 5). However, co-culturing with undifferentiated hMSCs did not alter the number of Cal27 and HLaC78 cells compared with that in monoculture cells (Fig. 5). Furthermore, there was no significant difference in the viability of HLaC78 cells after co-cultivation with osteo-hMSCs (Fig. 5).

Following the treatment of Cal27 and HLaC cells with media conditioned by hMSCs, increased cell viability was observed. Adipo-hMSC-CM treatment significantly enhanced tumor cell viability compared with that in cells treated with osteo-hMSC-CM (Fig. 6). There was no statistically significant difference in the cell count compared to the incubation with the control groups DMEM or hMSC-CM (Fig. 6).

Dot blot assay was used to investigate the profile of cytokine secretion in hMSCs, adipo-hMSCs and osteo-hMSCs. Due to the high expression level and their central role in tumour growth stimulation and inflammation, IL-6 was chosen for further detailed analysis. The secretion of IL-6 by adipo-hMSCs was markedly higher compared with that by osteo-hMSCs, but lower compared with in hMSCs (Fig. 7).

According to ELISA, IL-6 levels in the Cal27 cell culture supernatants were markedly increased after incubation with hMSC-CM and co-cultivation of Cal27 with hMSCs, compared with those in the supernatant of control Cal27 cells incubated with DMEM (Fig. 7F and G). In addition, comparably high levels of IL-6 were detected after incubation of Cal27 cells with hMSC-CM and in those co-cultured with osteo-hMSCs compared to Cal27 cells incubated with DMEM (Fig. 7F and G).

STAT3 activation at protein level was next evaluated by western blotting. Cal27 and HLaC78 cells were cultured with hMSC-CM, adipo-hMSC-CM and osteo-hMSC-CM. Furthermore to evaluate the influence of IL-6 on STAT3-activation, anti-IL-6 was added to hMSC-CM, adipo-hMSC-CM and osteo-hMSC-CM. The cultivation of Cal27 and HLaC78 cells with DMEM-EM served as the control. Markedly enhanced activation of STAT3 was observed after the treatment of Cal27 cells with CM compared with that in the control group (Fig. 8). However this was not observable for HLaC78 cells. The adipogenic or osteogenic differentiation had no influence on the level of STAT3-activation. The addition of anti-IL-6 reduced the STAT3 phosphorylation.